This article was downloaded by: [Emory University]On: 28 August 2015, At: 13:49Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,London, SW1P 1WG

Click for updates

Cell CyclePublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/kccy20

CTCF-dependent co-localization of canonical Smadsignaling factors at architectural protein binding sitesin D. melanogasterKevin Van Bortleac, Aidan J Petersonb, Naomi Takenakaa, Michael B O'Connorb & Victor GCorcesa

a Department of Biology; Emory University, Atlanta, GA USAb Department of Genetics; Cell Biology; and Development; University of Minnesota,Minneapolis, MN USAc Current affiliation: Department of Genetics; Stanford University, Stanford, CA USAAccepted author version posted online: 30 Jun 2015.

To cite this article: Kevin Van Bortle, Aidan J Peterson, Naomi Takenaka, Michael B O'Connor & Victor G Corces (2015) CTCF-dependent co-localization of canonical Smad signaling factors at architectural protein binding sites in D. melanogaster, CellCycle, 14:16, 2677-2687, DOI: 10.1080/15384101.2015.1053670

To link to this article: http://dx.doi.org/10.1080/15384101.2015.1053670

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

CTCF-dependent co-localization of canonicalSmad signaling factors at architectural protein

binding sites in D. melanogasterKevin Van Bortle1,#, Aidan J Peterson2, Naomi Takenaka1, Michael B O’Connor2, and Victor G Corces1,*

1Department of Biology; Emory University, Atlanta, GA USA; 2Department of Genetics; Cell Biology; and Development; University of Minnesota, Minneapolis, MN USA

#Current affiliation: Department of Genetics; Stanford University, Stanford, CA USA

Keywords: BMP, chromatin, epigenetics, growth factor, Transcription, TGF-b signaling

The transforming growth factor b (TGF-b) and bone morphogenetic protein (BMP) pathways transduce extracellularsignals into tissue-specific transcriptional responses. During this process, signaling effector Smad proteins translocateinto the nucleus to direct changes in transcription, but how and where they localize to DNA remain importantquestions. We have mapped Drosophila TGF-b signaling factors Mad, dSmad2, Medea, and Schnurri genome-wide in Kccells and find that numerous sites for these factors overlap with the architectural protein CTCF. Depletion of CTCF byRNAi results in the disappearance of a subset of Smad sites, suggesting Smad proteins localize to CTCF binding sites ina CTCF-dependent manner. Sensitive Smad binding sites are enriched at low occupancy CTCF peaks within topologicaldomains, rather than at the physical domain boundaries where CTCF may function as an insulator. In response toDecapentaplegic, CTCF binding is not significantly altered, whereas Mad, Medea, and Schnurri are redirected from CTCFto non-CTCF binding sites. These results suggest that CTCF participates in the recruitment of Smad proteins to a subsetof genomic sites and in the redistribution of these proteins in response to BMP signaling.

Introduction

Transforming growth factor b (TGF-b) and bone morphoge-netic protein (BMP) signaling effector proteins, called Smads,direct the transcriptional response to signaling pathways involvedin controlling cellular homeostasis, proliferation, differentiation,and apoptosis.1,2 In the canonical TGF-b signaling pathway, ser-ine-threonine kinase transmembrane receptors bind to extracellu-lar TGF-b ligands, and in turn phosphorylate receptor-regulatedSmad proteins, which are then able to form complexes that trans-locate into the nucleus to promote transcriptional activation andrepression.3 Several recent studies have focused attention onuncovering the chromatin determinants of Smad localization.Master regulatory transcription factors, which control the tran-scription of key cellular identity genes and are themselvesexpressed in a cell-type specific manner, were shown to direct thelocalization of BMP and Wnt signaling factors in haematopoieticprogenitor cells, and of TGF-b effector Smad proteins in embry-onic stem cells (ESCs).4,5 The co-localization of Smads atcell-type specific master transcription factor binding sites inmulti-potent cells provides an attractive model for how TGF-bsignaling events might produce diverse, tissue-specific responses.

In addition to master transcription factors, TRIM33, a multi-functional Smad-interacting protein that recognizes the dual

histone mark motif of H3K9me3 and H3K18ac, also creates aplatform for Smad localization in response to signaling events inESCs.6 Nevertheless, the mechanisms driving recruitment ofSmad proteins to DNA in non-stem cells remain largely unex-plored. Studies probing the mammalian Alzheimer amyloid pre-cursor protein (APP) gene promoter7 and the H19 imprintingcontrol region8 have identified sites in which Smad recruitmentto DNA is mediated by the architectural protein CTCF, previ-ously characterized for its ability to mediate long-range interac-tions and to function as an insulator.9 To what degree CTCFrecruits Smad proteins to DNA on a global scale, whether theseinteractions are related to the TGF-b response, and whetherCTCF-directed Smad localization is conserved in other organ-isms remains unknown.

CTCF and other architectural proteins establish high occu-pancy architectural protein binding sites (APBSs) at the bordersof topologically associating domains (TADs), which representhighly self-interacting regions of eukaryotic chromosomes.10–15

Both TADs and high occupancy APBSs appear to be largely tis-sue-invariant,14 suggesting most modular chromatin domains areconserved throughout development. However, intra-TAD inter-actions, which may be facilitated by low occupancy APBSs withindomains, often consist of enhancer-promoter and promoter-pro-moter interactions that are likely to be dynamically regulated in

*Correspondence to: Victor G Corces; Email vcorces@emory.eduSubmitted: 04/06/2015; Revised: 05/08/2015; Accepted: 05/16/2015http://dx.doi.org/10.1080/15384101.2015.1053670

www.tandfonline.com 2677Cell Cycle

Cell Cycle 14:16, 2677--2687; August 15, 2015; © 2015 Taylor & Francis Group, LLCREPORT

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

response to cell signaling events and between cell types to pro-duce cell-type specific transcription patterns.16-19 We have previ-ously demonstrated that Drosophila architectural proteins exhibitmoderate changes in genome-wide localization that correlatewith dynamic chromosome organization in response to the insectsteroid hormone 20-hydroxyecdysone,20 which is bound by thenuclear ecdysone receptor, a ligand-activated transcription factorthat in turn activates specific genes.21 However, to what degreearchitectural protein binding contributes to additional signalingevents remains poorly addressed.

Here we report the genome-wide landscape of the receptor-regulated Smad proteins Mothers against DPP (Mad) anddSmad2, the co-Smad Medea, and the co-repressor Schnurri inD. melanogaster. Indeed, we identify numerous sites in whichSmad binding co-localizes with the Drosophila homolog ofCTCF (dCTCF), and further demonstrate that dCTCF bindingis required for co-localization at a subset of sites. These data sup-port a role for CTCF as a conserved determinant of Smad locali-zation from Drosophila to humans. Finally, we demonstrate thatdynamic binding of Smad proteins in response to the Drosophilabone morphogenetic protein Decapentaplegic (DPP), a memberof the TGF-b superfamily, occurs in the context of dCTCF-inde-pendent binding sites and that dCTCF binding itself remainsunchanged. Together, these results suggest that whereas architec-tural protein occupancy is not dynamically regulated in responseto TGF-b, signaling events otherwise redirect Smad localizationto promote changes in transcription independent of CTCF.

Results

Genome-wide mapping of Drosophila Smad proteinsTGF-b superfamily signaling in D. melanogaster is tradition-

ally broken into 2 branches based on the ligand-receptor interac-tion and the class of receptor-regulated Smad proteins (R-Smads)that are subsequently phosphorylated.43 BMP ligands bind thetype I serine-threonine receptors Thickveins (Tkv) and Saxo-phone (Sax) to induce C-terminal phosphorylation of theR-Smad Mothers against DPP (Mad).44,45 TGF-b/Activinligands signal through splice isoforms of the Baboon receptor toactivate the R-Smad dSmad2 (Smad on X).46-48 The phosphory-lated forms of dSmad2 and Mad are able to form heterotrimericcomplexes with the co-Smad Medea, and upon which translocateinto the nucleus to direct changes in transcription.49 The Mad/Medea complex has been shown to target activation regulatoryelements in response to DPP,50 as well as repressive regulatoryelements to which the transcriptional repressor Schnurri isrecruited.51,52 The expression of several genes depends on Activinsignaling components,53-55 but dSmad2 regulatory elements havenot been identified. Therefore, in order to effectively capture thechromatin landscape of TGF-b signaling pathways, we carriedout ChIP-seq experiments against dSmad2, Mad, Medea, andSchnurri in the late-embryonic Drosophila hemocyte cell lineKc167 (Fig. 1, Supplemental Table 1).

Initial genome-wide analysis of Smad binding reveals extensiveoverlap between R-Smads and the co-Smad Medea (Mad overlap

with Medea – 85%; dSmad2 overlap with Medea – 81%;p < 0.0001, permutation test) as well as with the co-repressorSchnurri (Mad overlap with Schnurri – 67%; dSmad2 overlapwith Schnurri – 70%; p < 0.0001 permutation test) (Fig. 1A).Interestingly, nearly half of all binding sites for either Mad ordSmad2 also overlap with the distinct R-Smad (Mad overlap withdSmad2 – 52%; dSmad2 overlap with Mad – 47%; p < 0.0001,permutation test), suggesting a potential interaction betweenTGF-b/Activin and BMP signaling pathways at the genomiclevel. In support of this observation, recent studies have showncomplex formation between TGF-b and BMP Smad proteins atBMP response elements in human epithelial cells,56 suggestingcross-talk between TGF-b and BMP signaling factors may be aconserved phenomenon. Ligand stimulation of the TGF-b recep-tor Baboon can also induce phosphorylation of the BMP R-SmadMad,27 and mutation of the Activin R-Smad dSmad2 leads toaltered BMP pathway signaling,57 suggesting a potentially strongdegree of interaction between TGF-b and BMP signaling path-ways both upstream and downstream of nuclear translocation.Surprisingly, numerous Medea and Schnurri binding sites arepresent independently of either Mad or dSmad2, though a major-ity of Schnurri binding sites overlap with Medea (Fig. 1A), sug-gesting that additional regulatory factors may be involved inrecruiting Medea and Schnurri to DNA, or that Medea andSchnurri independently localize to specific regulatory elements.

Identification of genomic loci co-bound by Medea, Schnurri,and either R-Smad reveals even greater overlap between the BMPand TGF-b signaling factors Mad and dSmad2 (Fig. 1B; 69%for Mad; 66% for dSmad2; p < 0.0001, permutation test), sug-gesting both R-Smads co-localize with Medea and Schnurri atsimilar target sequences. The 1,220 overlapping Schnurri,R-Smad, Medea target sites bound by both Mad and dSmad2provide a high confidence list of signaling target regions com-monly identified in 4 independent ChIP-seq experiments that weconsider for further analysis. We hereafter refer to these sitesco-bound by dSmad2, Mad, Medea, and Schnurri as dSMMSmodules analogous to previously defined target elements.52

Visual inspection of high confidence dSMMS modules confirmsco-localization of these signaling proteins at well characterizedtarget genes. For example, brinker, a gene that encodes a nuclearrepressor that antagonizes DPP signaling by binding similarMad/Medea target sequences, is transcriptionally repressed byMad, Medea, and Schnurri in response to DPP signaling events.Accordingly, the brinker gene was recently shown to include sev-eral promoter modules targeted by Mad, Medea, and Schnurri.58

Although brinker is modestly expressed in Drosophila Kc167cells, our ChIP-seq data provide evidence that these dSMMSmodules are highly occupied in cell culture even prior to induc-tion with DPP (Fig. 1C), suggesting either a minimal level of sig-naling occurs in Kc167 cells or that Smad-mediated repression isregulated downstream of sequence binding.

dSMMS modules overlap Drosophila CTCF and otherarchitectural proteins

De novo motif analysis on dSMMS modules using MEME-ChIP59 identifies enrichment for 2 sequences that show

2678 Volume 14 Issue 16Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

similarity to putative Mad and Medeaconsensus sequences (Fig. 2). Previouscomparisons of specific Mad/Medea-binding elements suggest that Smadstarget GCCGNC as a consensus bind-ing sequence,60,61 and, similarly,MEME-ChIP identifies enrichmentfor GCYGSC at dSMMS modules(Fig. 2A). dSMMS modules are nearlyequally enriched for a GC rich con-sensus sequence with strong similarityto putative Medea binding sites,together suggesting that our ChIP-seqprofiles for dSmad2, Mad, Medea,and Schnurri provide an accuratemeans for identifying Smad-signalingresponse elements.

Interestingly, motif analysis alsoidentifies a consensus sequence previ-ously identified as being enriched atDrosophila CTCF binding sites thatoverlap with additional architecturalproteins, such as Boundary ElementAssociated Factor of 32 kDa (BEAF-32), Centrosomal Protein 190(CP190), Modifier of mdg4 (Mod(mdg4)), Chromator, and the cohesinand condensin complex kleisensubunits Rad21 and CAP-H2 respec-tively.10,11 In fact, we find significantmotif enrichment for DNA bindingarchitectural protein consensussequences targeted by dCTCF,BEAF-32, and CP190, whereassequences targeted by Suppressor ofHairy-wing (Su(Hw)) are depleted indSMMS modules (Fig. 2B). Thisresult indicates that genomic elementsbound by R-Smads Medea andSchnurri often occur in close spatialproximity with APBSs, and suggeststhat architectural proteins may playan important role in either creating an accessible chromatinlandscape for dSMMS occupancy, or may themselves directlyrecruit Smads to DNA.

Visualization of dSMMS modules overlapping APBSs illus-trates that ChIP-seq read densities for BMP and TGF-b Smadproteins tightly intersect those of dCTCF and other architecturalproteins (Fig. 2C). Genome-wide, 62% of dSMMS modules co-localize with dCTCF (p < 0.0001, permutation test), and amajority of modules also co-localize with additional architecturalproteins (Fig. 2D). These data, together with motif enrichmentprofiles for architectural proteins in Smad binding sites, suggestthat architectural proteins may play an important role in Smadlocalization.

dCTCF-dependent co-localization of Smad proteinsat APBSs

The strong overlap of BMP and TGF-b effectors withDrosophila CTCF suggests that CTCF-dependent recruitmentof Smad proteins may be an important, highly conservedgenome-wide phenomenon. In order to test whether Smadproteins target Drosophila APBSs in a dCTCF-dependentmanner, we repeated ChIP-seq experiments for Mad,dSmad2, and Medea in Kc167 cells depleted of dCTCF byRNAi.10,11 In all cases, disruption of dCTCF levels signifi-cantly perturbs the levels of Smad ChIP-seq read density at asubset of sites (Figs. 3A–C). For example, Mad occupancysignificantly decreases at more than 200 binding sites

Figure 1. Genome-wide mapping of BMP and TGF-b signaling proteins Mad, dSmad2, Medea, andSchnurri in Drosophila Kc167 cells. (A) Heatmap representation of ChIP-seq peak overlap between indi-vidual Smad proteins. Corresponding values represent percentage of total binding sites for factorsalong the horizontal axis that overlap peaks identified for factors along the vertical axis, ranging from 0(blue) to 100 (red). (B) Overlap of Schnurri-Mad-Medea modules individually identified for receptor-reg-ulated Smad proteins Mad or dSmad2. Overlap is statistically significant (p < 0.0001, permutation test).(C) Illustration of ChIP-seq experiments and Smad protein overlap at dSMMS regulatory elements pres-ent in previously characterized promoter-proximal modules of the brinker gene locus.

www.tandfonline.com 2679Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

(Fig. 3A), and these sites are significantly enriched for thedCTCF motif. Whereas fewer dSmad2 and Medea peaks aresignificantly downregulated using the same threshold (abso-lute M value cutoff of 1, p < 0.02), all Medea and dSmad2peaks identified as being lost after dCTCF RNAi indeed

overlap dCTCF (p < 0.0001, permutation test), suggestingloss of Smad binding is a direct consequence of dCTCFdepletion.

Strikingly, Mad, Medea, and dSmad2 binding sites that areaffected by dCTCF depletion are most often entirely lost after

Figure 2. dSMMS module motif enrichment and overlap at architectural protein binding sites. (A) Putative Mad and Medea motifs identified de novo byMEME-ChIP and the fraction of dSMMS modules containing the identified consensus sequence. Sequences are aligned with database motifs for Mad(top) and Medea (bottom) predicted based on previous DNA-binding experiments. Alignment and p-values defined by TOMTOM motif comparison tool.(B) Consensus sequence enrichment (log2(observed/expected frequency)) for Mad, Medea, and dCTCF/CP190 motifs identified de novo by MEME-ChIP(*) along with previously characterized DNA-binding architectural protein consensus sequences. (C) Example genomics viewer illustration of dSMMSmodule overlap at architectural protein binding sites at the Ribosomal Protein L30 (RpL30) gene promoter. (D) Percentage of dSMMS modules overlap-ping architectural protein binding sites for dCTCF, BEAF-32, CP190, Rad21, Cap-H2, Mod(mdg4), and Chromator. Comparison of observed overlapfrequencies (blue) with randomized shuffle control peaks (yellow) illustrates the strong enrichment of dSMMS modules at APBSs.

2680 Volume 14 Issue 16Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

Figure 3. RNAi depletion of Drosophila CTCF abrogates Smad localization to a subset of dCTCF binding sites. (A–C) MA plots depicting changes in ChIP-seq readdensity as a function of average peak read densities for Mad, Medea, and dSmad2 respectively. dCTCF RNAi and control ChIP-seq experiments were normalizedusing the MAnorm package for quantitative comparison of ChIP-seq data sets.39 An M value (log2 ratio of ChIP-seq read density) threshold cutoff of 1 (dotted line)and a p value cutoff of 0.02 (red) were chosen for consideration of significantly changing peaks. (D-E) Genomic viewer comparison of dCTCF RNAi (red) and controlChIP-seq experiments (black) for Mad, Medea, and dSmad2 respectively. In all cases, several significant ChIP-seq peaks overlapping dCTCF are lost in response toknockdown of dCTCF (blue arrow), whereas non-overlapping Smad binding sites remain comparatively unchanged.

www.tandfonline.com 2681Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

dCTCF RNAi (Figs. 3D-F), suggesting recruitment of Smadproteins to these loci is entirely dependent on dCTCF. Theseresults mirror the dependence of Smad localization to both theH19 imprinting control region and the Alzheimer b-amyloidprecursor protein proximal promoter region on human CTCF.However, despite the significant drop in Smad occupancy at asubset of sites in Drosophila Kc167 cells, ChIP-seq read densityfor Mad, Medea, and dSmad2 is not otherwise depleted at alldCTCF sites (Supplemental Fig. 1), suggesting dSMMSelements are differentially affected by, and perhaps differentiallydependent on, dCTCF.

dCTCF-dependent Smad binding occurs at low occupancyAPBSs within topological domains

Drosophila CTCF and other architectural proteins targetthousands of regulatory elements that play unique roles in shap-ing chromosome organization and transcription. For example,long-range interactions mediated by architectural proteins canfacilitate either active enhancer-promoter interactions or repres-sive Polycomb response element interactions, whereas sites boundby numerous architectural proteins are involved in establishingdiscrete topological domains and are capable of robust insulatorfunction for which these proteins were originally characterized.We therefore asked whether Smad binding sites that are sensitiveor insensitive to dCTCF RNAi occur within unique contexts.Here we compare Mad, Medea, and dSmad2 binding sites over-lapping dCTCF that are lost in response to dCTCF depletionwith overlapping dSMMS modules that do not change inresponse to RNAi.

Motif analysis reveals significant enrichment for the dCTCFcore consensus sequence at sites in which Mad, Medea, and/ordSmad2 are lost after dCTCF knockdown (Fig. 4A). In con-trast, Smad binding sites in which dCTCF co-localizationoccurs independently of dCTCF are enriched for the BEAF-32motif and consensus sequences known to be enriched at highoccupancy APBSs. Drosophila CTCF-independent, overlappingSmad peaks are also enriched for the putative Mad and Medeamotifs identified by MEME-ChIP, whereas sites sensitive todCTCF knockdown are depleted for Mad and Medea consensussequences. These differences in motif enrichment suggest poten-tially unique binding modes for Smad proteins at dCTCF-inde-pendent versus dCTCF-dependent peaks. Intuitively, Smadpeaks containing direct target sequences are not sensitive todCTCF depletion, whereas sites lacking direct binding motifsappear to be entirely dependent on dCTCF for recruitment tothese loci.

Enrichment of the BEAF-32 consensus sequence anddCTCF/CP190 motifs present at high occupancy APBSs alsosuggests that, in addition to dCTCF, other architectural proteinsmay provide some redundancy in recruiting Smad proteins toAPBSs. Comparison of DNase I hypersensitivity at dCTCF-sen-sitive and dCTCF-insensitive Smad peaks by DNase-seq in Dro-sophila Kc167 cells provides additional evidence that dCTCF-independent binding occurs at high occupancy APBSs. Forexample, whereas Smad peaks that are lost after dCTCF knock-down exhibit sharp but comparatively weaker DNase I

hypersensitivity, dCTCF-independent Smad binding sites arecharacterized by broad, robust DNase sensitivity commonlyobserved at high occupancy APBSs (Fig. 4B). Accordingly,dCTCF-independent Smad peaks overlap on average more archi-tectural proteins than dCTCF-dependent loci (Fig. 4C). Thoughthe difference in overlap does not reach statistical significance(p D 0.061, wilcoxon rank-sum test), this comparison draws onbinary peak identification by MACS and does not take intoaccount the relative ChIP-seq tag enrichment.

The observed differences in consensus sequence enrichment,DNase I hypersensitivity, and total architectural protein occupancyat differentially affected Smad binding sites suggest that dCTCF-dependent and dCTCF-independent Smad peaks are likely presentin unique chromosomal contexts. Integration of ChIP-seq withgenome-wide chromosomal interaction mapping has recentlyshown that interphase chromosomes are organized into discrete,self-interacting domains that are separated by high occupancyAPBSs.10 Accordingly, we find that Mad, Medea, and dSmad2peaks that overlap dCTCF but are not significantly affected bydCTCF depletion are enriched near the boundaries of topologicaldomains (Fig. 4D). In contrast, dCTCF-dependent Smad bindingsites do not predominantly localize near TAD borders, consistentwith the nature of low occupancy dCTCF binding sites residingwithin topological domains. Recent enhancer-trap assays suggestthat individual topological domains functionally limit the activityof enhancers to genes that reside within the same TAD.62 We spec-ulate that, whereas high occupancy APBSs mediate long-rangehigher order chromosomal domain organization, low occupancydCTCF binding sites may facilitate short-range enhancer-promoterinteractions relevant to transcription. However, to what degreedCTCF binding plays a role in the TGF-b signaling response hasnot been previously addressed.

dCTCF binding remains constant in response toDPP-induced signaling events

In order to assay whether dCTCF binding is influenced byTGF-b signaling events, we repeated ChIP-seq experiments fordCTCF as well as Mad, Medea, and Schnurri in response to theDrosophila bone morphogenetic protein DPP, a standardapproach for stimulating Mad phosphorylation in cell culture.Indeed, treatment with 30 nM DPP for 6 hr robustly activatesphosphorylation of Mad in Drosophila Kc167 cells, whereasphospho-Mad is undetectable in untreated control cells(Fig. 5A). Surprisingly, ligand-mediated activation of the BMPcascade does not dynamically alter the DNA-binding profile fordCTCF (Fig. 5B, less than 1% of CTCF binding sites increaseor decrease significantly), suggesting the architectural proteinoccupancy landscape remains relatively static. However, DPPdoes induce upregulation and downregulation of Mad (488 sitesdown, 161 sites up), Medea (9 sites down, 63 sites up), andSchnurri (109 sites down, 90 sites up) across the genome (Fig-s. 5C–E, Supplemental Fig. 2), suggesting Smad localization isdynamically altered in response to signal transduction.

We next performed gene ontology (GO) analysis on genesthat are in close spatial proximity to dSMMS modules and nearsites that increase or decrease in response to DPP (<4 kb from

2682 Volume 14 Issue 16Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

TSS). Whereas Mad, Medea, and Schnurri peaks that decreaseare enriched near genes associated with imaginal disc develop-ment, consistent with dSMMS module occupancy prior toinduction with DPP, dynamically upregulated binding sitesoccur in close spatial proximity to genes associated with neuro-genesis (Fig. 5F). Though DPP is known to influence neuronaldevelopment in a context-specific manner, whether these dynam-ics are at all meaningful for Kc167 cellular development

in vivo remains unknown. Nevertheless, comparison of upregu-lated Smad peaks with respect to dCTCF binding suggests thatthis signaling transcriptional response occurs in a non-dCTCFcontext. For example, upregulated Mad binding sites are depletedfor dCTCF, and co-localization of Medea and Schnurri upregu-lated peaks with dCTCF is no greater than expected by randomchance (Supplemental Fig. 3). DPP-induced downregulation ofSmad binding, on the other hand, does occur at dCTCF-binding

Figure 4. Differential motif enrichment, architectural protein occupancy, and chromosomal location of dCTCF-dependent versus independent Smadbinding sites. (A) Motif enrichment (log2(observed/expected frequency)) for architectural protein consensus sequences and putative Mad/Medea motifsat dCTCFRNAi lost (left) and dCTCFRNAi stable Smad binding sites. Comparison includes only Smad peaks which overlap dCTCF. (B) DNase-seq read densi-ties centered on dCTCF binding sites overlapping Mad, Medea, and dSmad2 that are stable (black) or lost in response to dCTCF RNAi (red). (C) APBS occu-pancy, determined as the number of overlapping MACS-called peaks for architectural proteins, at Smad binding sites that are stable (black) or lost afterdCTCF knockdown (red) (p D 0.061, wilcoxon rank-sum test). (D) Distance (kb) from topologically associating domain (TAD) boundaries, at which highoccupancy APBSs are highly enriched.10 dCTCFRNAi lost Smad peaks are most abundant within domains, whereas dCTCFRNAi stable peaks are enrichednear TAD boundaries (median distance from TAD borders: 14.9 kb and 9.9 kb respectively, p D 3.9e-7, wilcoxon rank-sum test).

www.tandfonline.com 2683Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

sites, many of which coincide with low occupancy dCTCFAPBSs, suggesting loss of Smad proteins occurs at architecturalprotein-bound regulatory elements within topological domains.The dynamic localization of Mad, Medea, and Schnurri fromdCTCF sites to non-dCTCF sites suggests that DPP signalingmay actively redirect Smad localization to unique chromatincontexts relevant to cell differentiation.

Discussion

TGF-b effector proteins have been shown to co-localize withmammalian CTCF in a CTCF-dependent manner at just 2 indi-vidual loci. We now extend this observation to Drosophila using agenome-wide approach, providing evidence that architecturalprotein CTCF and canonical Smad signaling proteins, bothhighly conserved from fly to humans, co-localize on a globalscale. We further uncover context-specific features in whichSmad localization is dependent or independent of CTCF

binding. Interestingly, our genome-wide analysis identifies Mad,dSmad2, Medea, and Schnurri binding to previously character-ized response elements even in the absence of DPP ligand, inwhich levels of phosphorylated Mad are undetectable. This sig-nal-independent clustering of signaling proteins suggests that thegenomic TGF-b signaling response is not as simple as regulatingbinary “off vs. on” states, dependent on phosphorylated Mad.However, our attempts to map the genomic landscape of phos-phorylated-Mad before and after DPP stimulation were unsuc-cessful, likely due to issues with currently available p-Madantibodies. Though we could not determine the role of phos-phorylation as a determinant in Mad localization, it is conceiv-able that phosphorylation of Mad might play a role in regulatingthe resident time of DNA-binding, the recruitment of additionalregulatory partners, or the ability to establish functional long-range interactions.

We find that Smad co-binding at dCTCF sites is sensitive todCTCF depletion at low occupancy dCTCF target sequences forwhich Smad consensus sequences are depleted, whereas high

Figure 5. Dynamic Smad localization in response to DPP activated phosphorylation of Mad. (A) Western blot analysis of phosphorylated Mad levelsbefore and after 6 h treatment with DPP (30 nM), using phospho-specific p-Mad antibody. Loading control staining of histone H3. (B) MA plot depictingchanges in dCTCF ChIP-seq read density as a function of average peak read density. DPP treatment and control ChIP-seq experiments were normalizedusing the MAnorm package for quantitative comparison of ChIP-seq data sets.39 An M value (log2 ratio of ChIP-seq read density) threshold cutoff of 1(dotted line) and a p value cutoff of 0.02 (upregulated peaks – pink; downregulated peaks – blue) were chosen for consideration of significantly changingpeaks. Only 40 peaks decrease and 7 peaks increase significantly in response to DPP. (C–E) Analogous MA plots for Mad, Medea, and Schnurri ChIP-seqexperiments before and after treatment with DPP. Number of significantly changing peaks: Mad 161 increasing, 488 decreasing; Medea: 63 increasing, 9decreasing; Schnurri: 90 increasing, 109 decreasing. (F) Gene ontology (GO) analysis for dSMMS modules and significantly changing peaks in response toDPP treatment. GO term enrichment was performed for closest genes (TSS) within 4 kb of Mad, Medea, and Schnurri binding sites before treatment(black), that decrease (blue), or increase (pink) significantly after treatment with DPP. Whereas decreasing Smad binding sites are enriched near genesassociated with imaginal disc development and signaling, increasing binding sites occur near genes involved in neurogenesis.

2684 Volume 14 Issue 16Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

occupancy dCTCF binding sites co-bound by additional archi-tectural proteins remain unaffected. The dCTCF-independentrecruitment of Smads to high occupancy APBSs suggests thatadditional architectural proteins may redundantly recruit Smads,or simply provide an accessible chromatin landscape to whichMad, Medea, and dSmad2 can associate. Nevertheless, dCTCF-dependent localization of Smad proteins to specific low occu-pancy elements is consistent with the CTCF-dependent nature ofSmad binding at both the APP and H19 promoters inhumans.7,8 We speculate that dCTCF-dependent Smad localiza-tion to low occupancy APBSs within topological domains mayrepresent regulatory elements involved in enhancer-promoterinteractions, whereas dCTCF-independent high occupancyAPBSs are involved in establishing higher-order chromosomeorganization. What role Smads might play in establishing ormaintaining such long-range interactions relevant to chromo-some architecture, or whether Smads and other transcription fac-tors simply localize to high occupancy APBSs due to chromatinaccessibility, remains difficult to address. However, we haverecently shown that high occupancy APBSs are distinct fromanalogous transcription factor hotspots, suggesting some level ofspecificity, most likely governed by protein-protein interactions,decides which factors can associate and where. Alternatively, theenrichment of ChIP-seq signal at high occupancy APBSs may, tosome degree, reflect indirect association via long-range interac-tions with regulatory elements directly bound by Smad proteins.This possibility raises a potential explanation for why SmadChIP signal is independent of dCTCF binding at high occu-pancy APBSs.

Surprisingly, DPP-activated phosphorylation of Mad doesnot lead to significant changes in dCTCF binding, whereasMad, Medea, and Schnurri levels increase at regulatory elementsaway from dCTCF. These results suggest that TGF-b signalingin Kc167 cells redirects Smad binding to genomic loci indepen-dent of architectural proteins, and that architectural proteinsmay facilitate binding of nuclear Smad proteins in the absenceof signaling. The complete loss of Smad ChIP signal at numer-ous dCTCF binding sites enriched for the core dCTCF consen-sus sequence nevertheless provides compelling evidence thatrecruitment of Smad proteins is directly governed by DrosophilaCTCF at a subset of binding sites. These results establishCTCF as an important determinant of Smad localization and,depending on the cell-type specific binding patterns of CTCF,suggest that CTCF might also influence the tissue-specific local-ization of Smad proteins analogous to master regulatory tran-scription factors in multi-potent stem cells.

Materials and Methods

Cell Culture and DPP TreatmentDrosophila Kc167 cells were grown in HyClone SFX cell cul-

ture medium. For treatment with DPP, cells were split overnightto a density of 0.5 £ 106 cells/mL and treated with 30 nM DPPthe follow morning. Recombinant Drosophila Decapentaplegicwas obtained from R&D systems (Cat#159-DP-20). DPP

treatment and control lysates were extracted after 6 hr incuba-tions. RNAi knockdown of Drosophila CTCF in cell culture wasconducted as per the Drosophila RNAi Screening Center(DRSC) protocol22 with the exception that dsRNA was addedevery day for 3 days and the cells were then collected on the 4th

day. Chromatin isolation from dCTCF depleted cell culturewere performed as part of previously published knockdownexperiments and described protocols.11 For a list of primersequences used for dCTCF RNAi as well as demonstration ofknockdown efficiency see Table S4 and Figure S7, respectively,at http://genome.cshlp.org/content/suppl/2012/08/01/gr.136788.111.DC1/Supplemental_materials.pdf.

Immunoprecipitation and Western AnalysisChromatin immunoprecipitation was performed as previously

described.23 Protein A Sepharose beads (GE Healthcare) werepre-washed in PBSMT and pre-incubated with 6 ml of rabbitpolyclonal anti-Mad,24 rabbit polyclonal anti-Medea,25 affinity-purified rabbit polyclonal anti-Schnurri,26 or rabbit polyclonalIgG (Santa Cruz sc-2027) for 1 hr prior to pull-down of shearedDNA. Affinity-purified sheep polyclonal anti-dSmad2 (R&Dsystems cat#AF7948) antibodies were pre-incubated with ProteinSepharose G beads (GE Healthcare). To generate sequencinglibraries, ChIP DNA was prepared for adaptor ligation by endrepair (End-It DNA End Repair Kit, Epicenter Cat# ER0720)and addition of “A” base to 30 ends (Klenow 30-50 exo–, NEBCat# M0212S). Illumina adaptors (Illumina Cat# PE-102-1001)were titrated according to prepared DNA ChIP sample concen-tration and ligated with T4 ligase (NEB Cat# M0202S). LigatedChIP samples were PCR-amplified using Illumina primers andPhusion DNA polymerase (NEB Cat# F-530L) and size selectedfor 200–300 bp by gel extraction. ChIP libraries were sequencedat the HudsonAlpha Institute for Biotechnology, using an Illu-mina HiSeq 2000.

For Western blotting, membranes were blocked in TBST(20 mM Tris, pH7.4, 150 mM NaCl, 0.05% Tween 20) with5% nonfat milk powder and incubated overnight with phospho-specific Mad antibodies.27 Membranes were washed 3 times withTBST and probed with secondary antibodies-conjugated to HRP(1:5000, Jackson ImmunoResearch Laboratories) for 1 hr. Afterthree more washes, the presence of different proteins was detectedusing SuperSignal West Pico/Dura Chemiluminescent substrate(Thermo Scientific).

ChIP-seq and reference dataIn addition to novel Mad, Medea, and Schnurri ChIP experi-

ments, architectural protein ChIP-seq data was obtained frompreviously published sources.10,11 Sequences were mapped to thedm3 genome with Bowtie 0.12.3,28 using default settings. Peakswere then called with MACS 1.4.0alpha229 using equal numbersof unique reads for IgG control and ChIP samples, using a pvalue cutoff of 1 £ 10¡10 For classification of overlapping bind-ing sites, MACS-identified peaks were intersected using publiclyavailable tools on Galaxy.30-32 For visualization, mappedsequence reads were loaded on to the Integrated GenomicsViewer.33,34 Previously published ChIP-seq data for Drosophila

www.tandfonline.com 2685Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

architectural proteins were obtained from GEO accessionsGSE3074020 and GSE36944.11 Raw DNase-seq in Kc167 cellswas obtained from published resources.35

Bioinformatics analysesDNA sequence motifs were identified by MEME-ChIP set

to identify 12 motifs and otherwise using default settings[28]. Sequences were aligned with database motifs and p-val-ues determined using the TOMTOM motif comparisontool.36 Comparison of APBSs with respect to Pol II-tran-scribed genes employed gene structure (TSSs) obtained usingthe UCSC genome browser.37,38 Significantly changingChIP-seq peaks in response to dCTCF RNAi and DPP sig-naling were determined using the MAnorm package forquantitative comparison of ChIP-seq data sets.39 An M value(log2 ratio of ChIP-seq read density) threshold cutoff of 1and a p value cutoff of 0.02 were chosen for considerationof significantly changing peaks. Enrichment profiles for Mad,Medea, Schnurri, and dSmad2 protein co-occurrence andoverlap with architectural proteins were defined as theobserved overlapping frequencies over expected frequenciesdetermined by shuffling datasets, while controlling for thenumber of peaks and start/stop location of peaks on eachchromosome. P-values were determined as the chance ofobserving an equal or greater co-occurrence across 100,000Monte Carlo Permutation tests. dSMMS modules weredefined as MACS-called R-smad peaks that overlap all 4 sig-naling proteins (dSmad2, Mad, Medea, and Schnurri), anddoes not require the presence of specific DNA sequences.Unless otherwise noted, read intensity plots were generatedby binning ChIP-seq reads into 100 bp bins and extractingread counts in bins surrounding described anchor points (eg.dSMMS modules or dCTCF peaks), and visualized usingJava Treeview.40 Gene Ontology analysis was performed by

running functional annotation tools on DAVID41 and bio-logical process (level 5) GO terms were subsequently summa-rized using REVIGO.42

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We are especially thankful to Drs. Lan Xu, Laurel Raftery, andMark Biggin for graciously sharing antibodies against Mad,Medea, and Schnurri. We thank the Genomic Services Lab at theHudsonAlpha Institute for Biotechnology for help in performingIllumina sequencing of ChIP-Seq samples.

Funding

This work was supported by US. Public Health ServiceAwards 5R01GM035463 to VGC and 5R01 GM095746 toMBO from the National Institutes of Health. The content issolely the responsibility of the authors and does not represent theofficial views of the National Institutes of Health.

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s website.

Accession Numbers

ChIP-seq data are deposited in NCBI’s Gene ExpressionOmnibus (GEO) (www.ncbi.nlm.nih.gov/geo) under accessionnumber GSE68654.

References

1. Liu F. Receptor-regulated Smads in TGF-beta signal-ing. Front Biosci 2003; 8:s1280-303;PMID:12957874; http://dx.doi.org/10.2741/1149

2. Ross S, Hill CS. How the Smads regulate transcription.Int J Biochem Cell Biol 2008; 40:383-408;PMID:18061509; http://dx.doi.org/10.1016/j.biocel.2007.09.006

3. Massague J. How cells read TGF-beta signals. Nat RevMol Cell Biol 2000; 1:169-78; PMID:11252892;http://dx.doi.org/10.1038/35043051

4. Trompouki E, Bowman TV, Lawton LN, Fan ZP, WuDC, DiBiase A, Martin CS, Cech JN, Sessa AK, LeblancJL, et al. Lineage regulators direct BMP and Wnt path-ways to cell-specific programs during differentiation andregeneration. Cell 2011; 147:577-89; PMID:22036566;http://dx.doi.org/10.1016/j.cell.2011.09.044

5. Mullen AC, Orlando DA, Newman JJ, Loven J, KumarRM, Bilodeau S, Reddy J, Guenther MG, DeKoter RP,Young RA. Master transcription factors determine cell-type-specific responses to TGF-beta signaling. Cell2011; 147:565-76; PMID:22036565; http://dx.doi.org/10.1016/j.cell.2011.08.050

6. Xi Q, Wang Z, Zaromytidou AI, Zhang XH, Chow-Tsang LF, Liu JX, Kim H, Barlas A, Manova-Todor-ova K, Kaartinen V, et al. A poised chromatin plat-form for TGF-beta access to master regulators. Cell2011; 147:1511-24; PMID:22196728; http://dx.doi.org/10.1016/j.cell.2011.11.032

7. Burton T, Liang B, Dibrov A, Amara F. Transforminggrowth factor-beta-induced transcription of the Alz-heimer beta-amyloid precursor protein gene involvesinteraction between the CTCF-complex and Smads.Biochem Biophys Res Commun 2002; 295:713-23;PMID:12099698; http://dx.doi.org/10.1016/S0006-291X(02)00725-8

8. Bergstrom R, Savary K, Moren A, Guibert S, HeldinCH, Ohlsson R, Moustakas A. Transforming growthfactor beta promotes complexes between Smad proteinsand the CCCTC-binding factor on the H19 imprint-ing control region chromatin. J Biol Chem 2010;285:19727-37; PMID:20427289

9. Phillips JE, Corces VG. CTCF: master weaver of thegenome. Cell 2009; 137:1194-211; PMID:19563753;http://dx.doi.org/10.1016/j.cell.2009.06.001

10. Van Bortle K, Nichols MH, Li L, Ong CT, Takenaka N,Qin ZS, Corces VG. Insulator function and topologicaldomainborder strength scalewitharchitecturalproteinoccu-pancy. Genome Biol 2014; 15:R82; PMID:24981874;http://dx.doi.org/10.1186/gb-2014-15-5-r82

11. Van Bortle K, Ramos E, Takenaka N, Yang J, Wahi J,Corces V. Drosophila CTCF tandemly aligns withother insulator proteins at the borders of H3K27me3domains. Genome research 2012; 22(11):2176-87;PMID:22722341

12. Hou C, Li L, Qin ZS, Corces VG. Gene density, tran-scription, and insulators contribute to the partition ofthe Drosophila genome into physical domains. Mol

Cell 2012; 48:471-84; PMID:23041285; http://dx.doi.org/10.1016/j.molcel.2012.08.031

13. Sexton T, Yaffe E, Kenigsberg E, Bantignies F, LeblancB, Hoichman M, Parrinello H, Tanay A, Cavalli G.Three-Dimensional Folding and Functional Organiza-tion Principles of the Drosophila Genome. Cell 2012;148(3):458-72; PMID:22265598; http://dx.doi.org/10.1016/j.cell.2012.01.010

14. Dixon JR, Selvaraj S, Yue F, Kim A, Li Y, Shen Y, HuM, Liu JS, Ren B. Topological domains in mammaliangenomes identified by analysis of chromatin interac-tions. Nature 2012; 485:376-80; PMID:22495300;http://dx.doi.org/10.1038/nature11082

15. Nora EP, Lajoie BR, Schulz EG, Giorgetti L, OkamotoI, Servant N, Piolot T, van Berkum NL, Meisig J, SedatJ, et al. Spatial partitioning of the regulatory landscapeof the X-inactivation centre. Nature 2012; 485:381-5;PMID:22495304; http://dx.doi.org/10.1038/nature11049

16. Handoko L, Xu H, Li G, Ngan CY, Chew E, SchnappM, Lee CW, Ye C, Ping JL, Mulawadi F, et al. CTCF-mediated functional chromatin interactome in pluripo-tent cells. Nat Genet 2011; 43:630-8;PMID:21685913; http://dx.doi.org/10.1038/ng.857

17. Phillips-Cremins JE, Sauria ME, Sanyal A, GerasimovaTI, Lajoie BR, Bell JS, Ong CT, Hookway TA, Guo C,Sun Y, et al. Architectural protein subclasses shape 3Dorganization of genomes during lineage commitment.

2686 Volume 14 Issue 16Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015

Cell 2013; 153:1281-95; PMID:23706625; http://dx.doi.org/10.1016/j.cell.2013.04.053

18. Seitan VC, Faure AJ, Zhan Y, McCord RP, Lajoie BR,Ing-Simmons E, Lenhard B, Giorgetti L, Heard E,Fisher AG, et al. Cohesin-based chromatin interactionsenable regulated gene expression within preexistingarchitectural compartments. Genome Res 2013; 23(12):2066-77; PMID:24002784

19. Sanyal A, Lajoie BR, Jain G, Dekker J. The long-rangeinteraction landscape of gene promoters. Nature 2012;489:109-13; PMID:22955621; http://dx.doi.org/10.1038/nature11279

20. Wood AM, Van Bortle K, Ramos E, Takenaka N,Rohrbaugh M, Jones BC, Jones KC, Corces VG. Regu-lation of chromatin organization and inducible geneexpression by a Drosophila insulator. Mol Cell 2011;44:29-38; PMID:21981916; http://dx.doi.org/10.1016/j.molcel.2011.07.035

21. Li TR, White KP. Tissue-specific gene expression andecdysone-regulated genomic networks in Drosophila.Dev Cell 2003; 5:59-72; PMID:12852852; http://dx.doi.org/10.1016/S1534-5807(03)00192-8

22. Armknecht S, Boutros M, Kiger A, Nybakken K,Mathey-Prevot B, Perrimon N. High-throughput RNAinterference screens in Drosophila tissue culture cells.Methods Enzymol 2005; 392:55-73; PMID:15644175

23. Bushey AM, Ramos E, Corces VG. Three subclasses ofa Drosophila insulator show distinct and cell type-spe-cific genomic distributions. Gen Dev 2009; 23:1338-50; PMID:19443682; http://dx.doi.org/10.1101/gad.1798209

24. Xu L, Yao X, Chen X, Lu P, Zhang B, Ip YT. Msk isrequired for nuclear import of TGF-{beta}/BMP-acti-vated Smads. J Cell Biol 2007; 178:981-94;PMID:17785517; http://dx.doi.org/10.1083/jcb.200703106

25. Sutherland DJ, Li M, Liu XQ, Stefancsik R, Raftery LA.Stepwise formation of a SMAD activity gradient duringdorsal-ventral patterning of the Drosophila embryo.Development 2003; 130:5705-16; PMID:14534137;http://dx.doi.org/10.1242/dev.00801

26. MacArthur S, Li XY, Li J, Brown JB, Chu HC, Zeng L,Grondona BP, Hechmer A, Simirenko L, Keranen SV,et al. Developmental roles of 21 Drosophila transcriptionfactors are determined by quantitative differences in bind-ing to an overlapping set of thousands of genomic regions.Genome Biol 2009; 10:R80; PMID:19627575; http://dx.doi.org/10.1186/gb-2009-10-7-r80

27. Peterson AJ, Jensen PA, Shimell M, Stefancsik R,Wijayatonge R, Herder R, Raftery LA, O’Connor MB.R-Smad competition controls activin receptor outputin Drosophila. PloS one 2012; 7:e36548;PMID:22563507; http://dx.doi.org/10.1371/journal.pone.0036548

28. Langmead B. Aligning short sequencing reads with Bow-tie. Current protocols in bioinformatics / editoral board,Andreas D Baxevanis; et al 2010; Chapter 11:Unit 11 7.

29. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS,Bernstein BE, Nusbaum C, Myers RM, Brown M, LiW, et al. Model-based analysis of ChIP-Seq (MACS).Genome Biol 2008; 9:R137; PMID:18798982; http://dx.doi.org/10.1186/gb-2008-9-9-r137

30. Goecks J, Nekrutenko A, Taylor J. Galaxy: a compre-hensive approach for supporting accessible, reproduc-ible, and transparent computational research in the lifesciences. Genome Biol 2010; 11:R86;PMID:20738864; http://dx.doi.org/10.1186/gb-2010-11-8-r86

31. Blankenberg D, Von Kuster G, Coraor N, Ananda G,Lazarus R, Mangan M, Nekrutenko A, Taylor J. Galaxy:a web-based genome analysis tool for experimentalists.Current protocols in molecular biology / edited by Freder-ick M Ausubel ; et al 2010; Chapter 19:Unit 199 0 1-21

32. Giardine B, Riemer C, Hardison RC, Burhans R, ElnitskiL, Shah P, Zhang Y, Blankenberg D, Albert I, Taylor J,et al. Galaxy: a platform for interactive large-scale genomeanalysis. Genome Res 2005; 15:1451-5; PMID:16169926;http://dx.doi.org/10.1101/gr.4086505

33. Robinson JT, Thorvaldsdottir H, Winckler W, Gutt-man M, Lander ES, Getz G, Mesirov JP. Integrativegenomics viewer. Nat Biotechnol 2011; 29:24-6;PMID:21221095; http://dx.doi.org/10.1038/nbt.1754

34. Thorvaldsdottir H, Robinson JT, Mesirov JP. Integra-tive Genomics Viewer (IGV): high-performance geno-mics data visualization and exploration. Briefbioinform 2013; 14:178-92; PMID:22517427; http://dx.doi.org/10.1093/bib/bbs017

35. Kharchenko PV, Alekseyenko AA, Schwartz YB, Min-oda A, Riddle NC, Ernst J, Sabo PJ, Larschan E,Gorchakov AA, Gu T, et al. Comprehensive analysis ofthe chromatin landscape in Drosophila melanogaster.Nature 2011; 471:480-5; PMID:21179089; http://dx.doi.org/10.1038/nature09725

36. Gupta S, Stamatoyannopoulos JA, Bailey TL, NobleWS. Quantifying similarity between motifs. GenomeBiol 2007; 8:R24; PMID:17324271; http://dx.doi.org/10.1186/gb-2007-8-2-r24

37. Kent WJ, Sugnet CW, Furey TS, Roskin KM, PringleTH, Zahler AM, Haussler D. The human genomebrowser at UCSC. Genome Res 2002; 12:996-1006;PMID:12045153; http://dx.doi.org/10.1101/gr.229102.

38. Meyer LR, Zweig AS, Hinrichs AS, Karolchik D, KuhnRM, Wong M, Sloan CA, Rosenbloom KR, Roe G,Rhead B, et al. The UCSC Genome Browser database:extensions and updates 2013. Nucleic Acids Res 2013;41:D64-9; PMID:23155063; http://dx.doi.org/10.1093/nar/gks1048

39. Shao Z, Zhang Y, Yuan GC, Orkin SH, Waxman DJ.MAnorm: a robust model for quantitative comparisonof ChIP-Seq data sets. Genome Biol 2012; 13:R16;PMID:22424423; http://dx.doi.org/10.1186/gb-2012-13-3-r16

40. Saldanha AJ. Java Treeview–extensible visualization ofmicroarray data. Bioinformatics 2004; 20:3246-8;PMID:15180930; http://dx.doi.org/10.1093/bioinformatics/bth349

41. Huang da W, Sherman BT, Zheng X, Yang J, Imami-chi T, Stephens R, Lempicki RA. Extracting biologicalmeaning from large gene lists with DAVID. Currentprotocols in bioinformatics / editoral board, Andreas DBaxevanis ; et al 2009; Chapter 13:Unit 13 1

42. Supek F, Bosnjak M, Skunca N, Smuc T. REVIGOsummarizes and visualizes long lists of gene ontologyterms. PloS one 2011; 6:e21800; PMID:21789182;http://dx.doi.org/10.1371/journal.pone.0021800

43. Peterson AJ, O’Connor MB. Strategies for exploringTGF-beta signaling in Drosophila. Methods 2014;68:183-93; PMID:24680699; http://dx.doi.org/10.1016/j.ymeth.2014.03.016

44. Brummel TJ, Twombly V, Marques G, Wrana JL,Newfeld SJ, Attisano L, Massague J, O’Connor MB,Gelbart WM. Characterization and relationship ofDpp receptors encoded by the saxophone and thickveins genes in Drosophila. Cell 1994; 78:251-61;PMID:8044839; http://dx.doi.org/10.1016/0092-8674(94)90295-X

45. Letsou A, Arora K, Wrana JL, Simin K, Twombly V,Jamal J, Staehling-Hampton K, Hoffmann FM, Gel-bart WM, Massague J, et al. Drosophila Dpp signalingis mediated by the punt gene product: a dual ligand-binding type II receptor of the TGF beta receptor fam-ily. Cell 1995; 80:899-908; PMID:7697720; http://dx.doi.org/10.1016/0092-8674(95)90293-7

46. Brummel T, Abdollah S, Haerry TE, Shimell MJ, Mer-riam J, Raftery L, Wrana JL, O’Connor MB. The Dro-sophila activin receptor baboon signals through dSmad2and controls cell proliferation but not patterning duringlarval development. Gen Dev 1999; 13:98-111;PMID:9887103; http://dx.doi.org/10.1101/gad.13.1.98

47. Jensen PA, Zheng X, Lee T, O’Connor MB. The Dro-sophila Activin-like ligand Dawdle signals preferentiallythrough one isoform of the Type-I receptor Baboon.Mech Dev 2009; 126:950-7; PMID:19766717; http://dx.doi.org/10.1016/j.mod.2009.09.003

48. Awasaki T, Huang Y, O’Connor MB, Lee T. Gliainstruct developmental neuronal remodeling through

TGF-beta signaling. Nat Neurosci 2011; 14:821-3;PMID:21685919; http://dx.doi.org/10.1038/nn.2833

49. Gao S, Steffen J, Laughon A. Dpp-responsive silencersare bound by a trimeric Mad-Medea complex. J BiolChem 2005; 280:36158-64; PMID:16109720; http://dx.doi.org/10.1074/jbc.M506882200

50. Weiss A, Charbonnier E, Ellertsdottir E, Tsirigos A,Wolf C, Schuh R, Pyrowolakis G, Affolter M. A con-served activation element in BMP signaling duringDrosophila development. Nat Struct Mol Biol 2010;17:69-76; PMID:20010841; http://dx.doi.org/10.1038/nsmb.1715

51. Muller B, Hartmann B, Pyrowolakis G, Affolter M,Basler K. Conversion of an extracellular Dpp/BMPmorphogen gradient into an inverse transcriptional gra-dient. Cell 2003; 113:221-33; PMID:12705870;http://dx.doi.org/10.1016/S0092-8674(03)00241-1

52. Pyrowolakis G, Hartmann B, Muller B, Basler K,Affolter M. A simple molecular complex mediateswidespread BMP-induced repression during Drosoph-ila development. Dev Cell 2004; 7:229-40;PMID:15296719; http://dx.doi.org/10.1016/j.devcel.2004.07.008

53. Zheng X, Wang J, Haerry TE, Wu AY, Martin J,O’Connor MB, Lee CH, Lee T. TGF-beta signalingactivates steroid hormone receptor expression duringneuronal remodeling in the Drosophila brain. Cell2003; 112:303-15; PMID:12581521; http://dx.doi.org/10.1016/S0092-8674(03)00072-2

54. Gibbens YY, Warren JT, Gilbert LI, O’Connor MB.Neuroendocrine regulation of Drosophila metamor-phosis requires TGFbeta/Activin signaling. Develop-ment 2011; 138:2693-703; PMID:21613324; http://dx.doi.org/10.1242/dev.063412

55. Chng WB, Sleiman MS, Schupfer F, Lemaitre B.Transforming Growth Factor beta/Activin SignalingFunctions as a Sugar-Sensing Feedback Loop to Regu-late Digestive Enzyme Expression. Cell Rep 2014;9:336-48; PMID:25284780; http://dx.doi.org/10.1016/j.celrep.2014.08.064

56. Gronroos E, Kingston IJ, Ramachandran A, RandallRA, Vizan P, Hill CS. Transforming growth factor betainhibits bone morphogenetic protein-induced tran-scription through novel phosphorylated Smad1/5-Smad3 complexes. Mol Cell Biol 2012; 32:2904-16;PMID:22615489; http://dx.doi.org/10.1128/MCB.00231-12

57. Peterson AJ, O’Connor MB. Activin receptor inhibi-tion by Smad2 regulates Drosophila wing disc pattern-ing through BMP-response elements. Development2013; 140:649-59; PMID:23293296; http://dx.doi.org/10.1242/dev.085605

58. Yao LC, Phin S, Cho J, Rushlow C, Arora K, WarriorR. Multiple modular promoter elements drive gradedbrinker expression in response to the Dpp morphogengradient. Development 2008; 135:2183-92;PMID:18506030; http://dx.doi.org/10.1242/dev.015826

59. Machanick P, Bailey TL. MEME-ChIP: motif analysisof large DNA datasets. Bioinformatics 2011; 27:1696-7; PMID:21486936; http://dx.doi.org/10.1093/bioinformatics/btr189

60. Xu X, Yin Z, Hudson JB, Ferguson EL, Frasch M.Smad proteins act in combination with synergistic andantagonistic regulators to target Dpp responses to theDrosophila mesoderm. Gen Dev 1998; 12:2354-70;PMID:9694800; http://dx.doi.org/10.1101/gad.12.15.2354

61. Kusanagi K, Inoue H, Ishidou Y, Mishima HK, Kawa-bata M, Miyazono K. Characterization of a bone mor-phogenetic protein-responsive Smad-binding element.Mol Biol Cell 2000; 11:555-65; PMID:10679014;http://dx.doi.org/10.1091/mbc.11.2.555

62. Symmons O, Uslu VV, Tsujimura T, Ruf S, Nassari S,Schwarzer W, Ettwiller L, Spitz F. Functional and topo-logical characteristics of mammalian regulatory domains.Genome Res 2014; 24:390-400; PMID:24398455;http://dx.doi.org/10.1101/gr.163519.113

www.tandfonline.com 2687Cell Cycle

Dow

nloa

ded

by [

Em

ory

Uni

vers

ity]

at 1

3:49

28

Aug

ust 2

015